Think you’re too small for GenAI? Think Again!

If you’ve been to a contact center trade show this year, you’ve probably seen a swath of new vendors touting their Generative AI wares. Also, the traditional vendors have gone all-in and have slapped “We do AI” stickers on their existing products or roadmaps, with a varying degree of completeness. 

And companies who have deployed these technologies have reported incredible success:

News outlets like The Australian reported earlier this year that Commonwealth Bank of Australia uses AI to handle approximately 50,000 inquiries per day through messaging and live chats, have reduced wait times 40% year over year, while other news outlets have mused at a world where they replace all of their human customer service agents.

Reuters reported that Verizon’s use of Generative AI is predicting the reasons for 80% of their 170 million annual customer calls, allowing better agent matching, driving retention. 

A paper published by ArXiv (by Cornell University) explained how a LLM-based knowledge chatbot called “Ask me Anything” has reduced handle time by Comcast customer service agents by 10%. 

So yes - these outcomes are amazing! And they’re also in the Enterprise market segment, where there are traditionally well-funded development teams working on the bleeding edge of technology. But what about the rest of us? Salesforce research indicates that only 24% of customer service professionals report using genAI in their work, and only 15% are planning to adopt it in the future. With other studies (Statista 2024) revealing that 60% of contact centers using generative AI reported improved customer effort scores, why aren’t more companies using it? 

I could muse that many contact centers are still bogged down by their legacy systems, lack a good data readiness strategy, or are just overwhelmed by the perceived complexity of implementing these new technologies. There’s also the fear of change and organizational inertia - especially if past technology investments haven’t delivered what it said on the label. I can imagine that this is like looking up a mountain with no clue on how to climb it… lightning strikes, and it starts raining. 

So, how can a small customer service operation progressively start a journey into Generative AI? Here’s my take:

• Find a small use case - or a few small use cases - that still have humans in the loop. For instance, the Local Measure platform, powered by Amazon Connect, has a growing handful of agent assistance features that rely on industry standard foundational models and are underpinned by Amazon Bedrock. In all of our use cases, an agent can edit the outputs - and are required to double check them before they’re sent to a customer. This low-risk approach is a good way to get some benefits quickly - such as after call work summaries, real-time messaging translation, entity extraction for form filling, and spelling, grammar, and smart composing. This strategy does not require you to have any data repositories, requires no code, and is completely customizable in a simple but powerful user interface. 

• Start to pull together a solid knowledge strategy. Generative AI is underpinned by knowledge, and if you are using a dozen spreadsheets in a shared drive and yellowing-from-age policy documents thumb-tacked to the wall in your agent’s workspace, you’re going to have a bad time. This may take some time, but take technology out of it - your agents are going to benefit as well.

• Use a contact center platform that generates a ton of data and allows you to consume it easily. The second pillar to generative AI strategy is to personalize the output using analyzed customer data. Customer purchase history, preferences, sentiment, previous conversations. All things that LLMs are incredible at analyzing very quickly. And for that, I recommend a platform like Amazon Connect - which generates a cornucopia of data in industry standard formats and doesn’t charge you a connector, doesn’t lock it up somewhere you don’t have access to it, and allows you to use it however you’d like (it is your data, isn’t it?).

• And finally, CX observability is crucial in this space. You’ve got conversations going on between non-human agents and customers. What’s going on in them? What happens when one gets stuck? Are your customers happy and getting their needs fulfilled? What context are you passing to your agents if one is required to intervene? I don’t have any advice here as this is a place where the technology is rapidly evolving - best practices are still taking shape. Keep a close eye on emerging solutions, pilot new tools in smaller use cases, and watch industry thought leaders and technology providers.

For a real-life example, let’s look at one of our customers—a small but rapidly growing entertainment startup. They started out using just one of our agent assist capabilities: entity extraction for form filling. Within a few hours—and without needing any IT resources—they cut down on repeat calls caused by agent errors by two-thirds. From there, they added more of our features, including after call summaries and our smart composer, and combined them with Amazon Connect’s own AI-driven tools for quality management. All of this allowed them to scale their customer base without adding staff at the same pace. In fact, they’ve reduced their growth-adjusted headcount by 80%. Their next focus? Knowledge and data. Even at this early stage of their company’s journey, their Generative AI strategy isn’t standing still—it’s growing right alongside their business needs.

Look, every company wants to deliver good customer service at the lowest possible price. And technology advancements have finally delivered on the promise. But remember, integrating Generative AI isn’t just about chasing the hottest buzzwords - it’s about taking a measured and incremental approach. Start small. Prioritize data readiness. Measure often. Refine your strategy. Thinking and executing like that, even the smallest contact centers can achieve results that were previously possible only by the largest enterprises. Your journey may look different than Verizon’s or Comcast’s, but the end goal - an efficient, engaging, and effortless customer experience - can be yours too.